Exploring a Hole Filling Technique in Reverse Engineering Domain
-
摘要: 对于散乱点云模型上的大面积、跨面孔洞,逆向软件往往难以修补。为了提高孔洞修补精度、获得完整的点云模型,提出了对手受惩罚竞争学习算法(Rival penalized competitive learning,RPCL)和模糊C均值聚类算法(Fuzzy C-means,FCM)相结合的综合改进径向基函数神经网络(RBF)算法,建立了基于改进算法的点云孔洞修补模型,并以挖掘机斗齿和汽车模型为研究对象,利用RPCL-FCM-RBF联合算法对不同特征的点云孔洞进行了修补研究。结果表明,该算法在很大程度上提高了点云孔洞的修补精度,其补洞效果远优于逆向软件。而且,较之传统的RBF神经网络,该方法所建模型具有更高的预测精度、能够有效地调整洞口缺失数据、实现点云孔洞的精确修复,实用性强。
-
关键词:
- 径向基函数神经网络(RBF) /
- 对手受惩罚竞争学习算法(RPCL) /
- 模糊C均值聚类算法(FCM) /
- 孔洞修补 /
- MATLAB
Abstract: For large holes with significant curvature change in scattered point clouds, the reverse engineering software usually has failure. In order to build up a complete point cloud model with a higher accuracy of hole repairing, an improved radial basis function (RBF) networks algorithm was exploited, where the rival penalized competitive learning (RPCL) algorithm was combined with the fuzzy clustering means (FCM) to perfect the traditional RBF networks, and the hybrid RPCL-FCM-RBF algorithm was then examined by applying it to two different scattered point clouds from scanned objects:a bucket of an excavator and a car. The experimental results indicate that the suggested method shows better hole-filling performance than the software and that the proposed algorithm has a superior prediction capability compared with the conventional RBF algorithm. The favorable fidelity and efficiency make it a promising candidate for many practical applications. -
[1] Quinsat Y, Lartigue C. Filling holes in digitized point cloud using a morphing-based approach to preserve volume characteristics[J]. The International Journal of Advanced Manufacturing Technology, 2015,81(1-4):411-421 [2] 王小超,曹俊杰,刘秀平,等.波前法在三角网格孔洞修补中的应用[J].计算机辅助设计与图形学学报,2011,23(6):1048-1054 Wang X C, Cao J J, Liu X P, et al. Advancing front method in triangular meshes hole-filling application[J]. Journal of Computer-Aided Design & Computer Graphics, 2011,23(6):1048-1054(in Chinese) [3] 刘咏梅,李凤霞,雷正朝,等.基于边界特征增长的孔洞修补算法[J].系统仿真学报,2014,26(9):1916-1921 Liu Y M, Li F X, Lei Z C, et al. New hole filling algorithm based on boundary-feature growing[J]. Journal of System Simulation, 2014,26(9):1916-1921(in Chinese) [4] Centin M, Pezzotti N, Signoroni A. Poisson-driven seamless completion of triangular meshes[J]. Computer Aided Geometric Design, 2015,35-36:42-55 [5] Yang L, Yan Q G, Xiao C X. Shape-controllable geometry completion for point cloud models[J]. The Visual Computer:International Journal of Computer Graphics, 2017,33(3):385-398 [6] Yumer M E, Kara L B. Surface creation on unstructured point sets using neural networks[J]. Computer-Aided Design, 2012,44(7):644-656 [7] 晏海平,吴禄慎,陈华伟.基于径向基函数的散乱点云孔洞修复算法[J].计算机工程与设计,2014,35(4):1253-1257 Yan H P, Wu L S, Chen H W. Hole repairing algorithm in scattered point cloud based on radial basis function[J]. Computer Engineering and Design, 2014,35(4):1253-1257(in Chinese) [8] 王春香,陈浩宏,孔艳彬,等.RBF神经网络在斗齿散乱点云漏洞修补中的应用[J].机械科学与技术,2014,33(10):1479-1483 Wang C X, Chen H H, Kong Y B, et al. Applications of RBF neural network in mending the 3D incomplete point clouds of different excavator bucket teeth[J]. Mechanical Science and Technology for Aerospace Engineering, 2014,33(10):1479-1483(in Chinese) [9] Niros A D, Tsekouras G E. A novel training algorithm for RBF neural network using a hybrid fuzzy clustering approach[J]. Fuzzy Sets and Systems, 2012,193:62-84 [10] 耿志强,陈杰,韩永明.基于模糊RBF神经网络的乙烯装置生产能力预测[J].化工学报,2016,67(3):812-819 Geng Z Q, Chen J, Han Y M. Ethylene plants production capacity forecast based on fuzzy RBF neural network[J]. CIESC Journal, 2016,67(3):812-819(in Chinese) [11] Wang Z O, Zhu T. An efficient learning algorithm for improving generalization performance of radial basis function neural networks[J]. Neural Networks, 2000,13(4-5):545-553 [12] 苏义鑫,夏慧雯.用于风电功率预测的RPCL优化神经网络模型[J].北京工业大学学报,2016,42(5):674-678 Su Y X, Xia H W. Rival penalized competitive learning-based neural network model for wind power forecasting[J]. Journal of Beijing University of Technology, 2016,42(5):674-678(in Chinese) [13] Czarnowski I, Jedrzejowicz P. Kernel-based fuzzy C-means clustering algorithm for RBF Network initialization[C]//Czarnowski I, Caballero A M, Howlett R J, et al. Intelligent Decision Technologies 2016. Cham:Springer, 2016:337-347 [14] Ngo H T M, Lee W S. Feature-First hole filling strategy for 3D meshes[C]//Csurka G, Kraus M, Mestetskiy L, et al. Computer Vision, Imaging and Computer Graphics. Theory and Applications. Berlin, Heidelberg:Springer, 2013:53-68 [15] Rouhani M, Javan D S. Two fast and accurate heuristic RBF learning rules for data classification[J]. Neural Networks, 2016,75:150-161 [16] Halali M A, Azari V, Arabloo M, et al. Application of a radial basis function neural network to estimate pressure gradient in water-oil pipelines[J]. Journal of the Taiwan Institute of Chemical Engineers, 2016,58:189-202 [17] Liao G L, Du L, Su L, et al. Using RBF networks for detection and prediction of flip chip with missing bumps[J]. Microelectronics Reliability, 2015,55(12):2817-2825 [18] Borghese A, Capraro W. Rival-penalized competitive clustering:a study and comparison[C]//Apolloni B, Bassis S, Esposito A, et al. Neural Nets and Surroundings. Berlin, Heidelberg:Springer, 2013:11-20 [19] 朱绍军.模型自动选择聚类算法的研究与应用[D].浙江宁波:宁波大学,2014 Zhu S J. Research and applications of clustering algorithms with the model selection ability[D]. Zhejiang Ningbo:Ningbo University, 2014(in Chinese) [20] 张永志,董俊慧.基于模糊C均值聚类的模糊RBF神经网络预测焊接接头力学性能建模[J].机械工程学报,2014,50(12):58-64 Zhang Y Z, Dong J H. Modeling fuzzy RBF neural network to predict of mechanical properties of welding Joints based on fuzzy C-means cluster[J]. Journal of Mechanical Engineering, 2014,50(12):58-64(in Chinese)
点击查看大图
计量
- 文章访问数: 167
- HTML全文浏览量: 25
- PDF下载量: 10
- 被引次数: 0